Poster #001, UCSF Benioff Children's Hospital Oakland
Using AI-based Image Analysis to Identify the Differentiation Efficiency of iPSCinduced Hepatocyte-like cell (iHep)
Mentors: Yuanyuan Qin, PhD and Marisa Medina, PhD
The prevalence of liver diseases is rising as a global burden. It is estimated that the global
liver-related mortality rate is about 2 million patients per year. To address this, researchers
have begun using hepatocyte-like cells differentiated from stem cells as tools in
personalized medicine applications. To authenticate hepatocyte presence from iPSC
differentiation, albumin, a liver-specific protein secreted into the medium, is used as an
indicator. Albumin is usually measured by ELISA, which is expensive and requires excessive
time and labor. Mature hepatocytes are polygonal cells with well-defined borders. Based
on their unique morphology, we hypothesize that with the use of bright field imaging, the
morphology of cells can be used to determine the differentiation efficiency of iPSCs to
iHeps.
Throughout the iPSC-iHep differentiation stages, daily cell images were taken to track
progress. We collected medium on Day 25 and measured albumin levels by ELISA. We also
quantified albumin expression using a fluorescence-labeled anti-albumin antibody for livecell imaging. We found that when the cells started to portray as iHeps, they also began to
biochemically act like hepatocytes, demonstrated through the presence of albumin
secretion. Additionally, a higher percentage of iHeps identified by the bright-field imaging is
correlated with higher albumin secretion in the medium and fluorescence intensity. We
also found that the bright-field images could predict the differentiation efficiency at an
earlier stage than Day 25. Our study demonstrated that the morphological characteristics
of cells can be used to indicate the presence of hepatocytes. Using this method, we can
simplify the authentication of iHeps earlier in differentiation and contribute to ongoing
studies to address the rising burden of liver diseases.